Cross-modal Vision-Language Retrieval on MSLS (val)
38R@1La-CLIP
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| La-CLIPAugmentation=Language-augmented (La-)2026.02 | 38 | 58.4 | 68.2 | 77.2 | |
| La-BLIPAugmentation=Language-augmented (La-)2026.02 | 38 | 60.5 | 68.4 | 75.4 | |
| La-SigLIP-V2Augmentation=Language-augmented (La-)2026.02 | 35.7 | 60.5 | 69.6 | 76.2 | |
| La-EVA-V2Augmentation=Language-augmented (La-)2026.02 | 32.8 | 52.8 | 62.4 | 70.9 | |
| EVA-CLIP-V2Augmentation=None2026.02 | 3.8 | 9.2 | 13 | 19.1 | |
| SigLIP-V2Augmentation=None2026.02 | 2.6 | 6.5 | 9.7 | 13.1 | |
| CLIPAugmentation=None2026.02 | 2.3 | 6.2 | 10.4 | 14.5 | |
| BLIPAugmentation=None2026.02 | 1.6 | 5.9 | 8.2 | 11.6 |